 Oh, Brad's attacking science, folks. No, no, exact count per interval. Interobserver agreement, IOA. First off, interobserver agreement is not interobserver reliability. We'll deal with that later. Interobserver agreement, exact count method. You have a series of behaviors that occur in intervals. You've broken your time and piece up into intervals. You have two observers. So how many times did behavior occur in this interval for person one? And how many times did behavior occur for this person in interval one? If those two numbers match, you've got it right. That's exact count. So it could be occurrence or non-occurrence. So if the behavior did not occur in the second interval and this person said that it did not occur, that's also an agreement. It's an exact count agreement. So if they saw three and they saw two, that doesn't count. So you take the number of intervals overall, where exact agreement between the two individuals occurred, and divide that by the total number of intervals. And you end up with a percentage, a measure of interobserver agreement. It actually speaks to believability about what happened in the observation setting. So again, the number of intervals where the agreement was identical divided by the total number of agreement. It's hyper stringent. It's really strict. That's it.